The Evolution of Reciprocal Altruism

Transcription

The Evolution of Reciprocal Altruism
The Evolution of Reciprocal Altruism
Author(s): Robert L. Trivers
Reviewed work(s):
Source: The Quarterly Review of Biology, Vol. 46, No. 1 (Mar., 1971), pp. 35-57
Published by: The University of Chicago Press
Stable URL: http://www.jstor.org/stable/2822435 .
Accessed: 20/10/2012 18:02
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact [email protected].
.
The University of Chicago Press is collaborating with JSTOR to digitize, preserve and extend access to The
Quarterly Review of Biology.
http://www.jstor.org
THE EVOLUTION OF RECIPROCAL ALTRUISM
BY
L.
ROBERT
TRIVERS
HarvardUniversity,
BiologicalLaboratories,
Cambridge,Mass. 02138
ABSTRACT
A model is presented to account for the natural selection of what is termed reciprocally altruistic behavior. The model shows how selection can operate -against the
cheater (non-reciprocator)in the system. Three instances of altruistic behavior are
discussed, the evolution of which the model can explain: (1) behavior involved in
cleaning symbioses; (2) warning cries in birds: and (3) human reciprocal altruism.
Regarding human reciprocal altruism, it is shown that the details of the psychological system that regulates this altruism can be explained by the model. Specifically,friendship,dislike, moralisticaggression,gratitude,sympathy,trust,suspicion,
aspects of guilt, and some forms of dishonestyand hypocrisycan be
trustworthiness,
explained as important adaptations to regulate the altruisticsystem. Each individual
human is seen as possessing altruistic and cheating tendencies, the expression of
which is sensitive to developmental variables that were selected to set the tendencies
at a balance appropriate to the local social and ecological environment.
A
INTRODUCTION
LTRUISTIC behavior can be definedas behavior that benefitsanotherorganism,not closelyrelated,
while being apparentlydetrimental to the organism performing
the behavior, benefitand detrimentbeing defined in termsof contributionto inclusive fitness. One human being leaping into water,at
some danger to himself,to save another distantly related human from drowning may be
said to display altruisticbehavior. If he were
to leap in to save his own child, the behavior
would not necessarilybe an instance of "al-truism";he may merelybe contributingto the
survivalof his own genes investedin the child.
Models that attemptto explain altruisticbehavior in termsof natural selectionare models
designed to take the altruismout of altruism.
For example, Hamilton (1964) has demonstrated that degree of relationship is an important parameterin predictinghow selection
will operate, and behavior which appears
altruistic may, on knowledge of the genetic
relationships of the organisms involved, be
explicable in termsof natural selection: those
genes being selectedfor thatcontributeto their
own perpetuation,regardlessof which individual the genes appear in. The term "kin selection" will be used in this paper to cover instancesof this type-that is, of organismsbeing
selectedto help theirrelativelyclose kin.
The model presented here is designed to
show how certain classes of behavior convenientlydenoted as "altruistic" (or "reciprocally altruistic")can be selected for even when
the recipient is so distantlyrelated to the organism performingthe altruisticact that kin
selection can be ruled out. The model will
apply, for example, to altruisticbehavior between membersof differentspecies. It will be
argued that under certain conditions natural
selection favors these altruistic behaviors because in the long run theybenefitthe organism
them.
performing
THE
MODEL
One human being savinganother,who is not
closely related and is about to drown, is an
instance of altruism. Assume that the chance
of the drowningman dying is one-half if no
one leaps in to save him, but that the chance
thathis potentialrescuerwill drownif he leaps
in to save him is much smaller, say, one in
35
36
TH
E QUARTERLY
REVIEW OF BIOLOGY
twenty. Assumethat the drowningman always
drowns when his rescuer does and that he is
always saved when the rescuersurvivesthe rescue attempt. Also assume that the energycosts
involvedin rescuingare trivialcomparedto the
survival probabilities. Were this an isolated
event, it is clear that the rescuer should not
bother to save the drowningman. But if the
drowningman reciprocatesat some futuretime,
and if the survival chances are then exactly
reversed,it will have been to the benefitof each
participantto have riskedhis life forthe other.
Each participant will have traded a one-half
chance of dying for about a one-tenthchance.
If we assume that the entire population is
sooner or later exposed to the same risk of
drowning,the two individuals who risk their
lives to save each other will be selected over
those who face drowningon their own. Note
that the benefitsof reciprocitydepend on the
unequal cost/benefitratio of the altruisticact,
that is, the benefitof the altruisticact to the
recipientis greaterthan the cost of the act to
the performer,cost and benefitbeing defined
here as the increase or decrease in chances of
the relevant alleles propagating themselvesin
the population. Note also that,as defined,the
benefitsand costs depend on the age of the
altruist and recipient (see Age-dependent
changesbelow). (The odds assigned above may
not be unrealistic if the drowning man is
drowningbecause of a cramp or if the rescue
can be executed by extending a branch from
shore.)
Why should the rescued individual bother to
reciprocate?Selection would seem to favorbeing saved fromdrowningwithoutendangering
oneself by reciprocating. Why not cheat?
("Cheating" is used throughout this paper
solely for convenience to denote failure to
reciprocate;no conscious intent or moral connotationis implied.) Selectionwill discriminate
against the cheaterif cheatinghas later adverse
affectson his life which outweighthe benefitof
not reciprocating.This may happen if the altruistrespondsto the cheatingby curtailingall
futurepossible altruisticgesturesto this individual. Assumingthat the benefitsof these lost
altruisticacts outweigh the costs involved in
reciprocating, the cheater will be selected
against relative to individuals who, because
neither cheats, exchange many altruisticacts.
[VOLUME
46
This argumentcan be made precise. Assume
there are both altruistsand non-altruistsin a
population of size N and that the altruistsare
characterizedby the fact that each performs
altruisticacts when the cost to the altruist is
well below the benefitto the recipient,where
cost is defined as the degree to which the behavior retardsthe reproductionof the genes of
the altruistand benefitis the degree to which
the behavior increasesthe rate of reproduction
of the genes of the recipient. Assume that the
altruisticbehavior of an altruist is controlled
by an allele (dominant or recessive),a2, at a
given locus and that (for simplicity)there is
only one alternativeallele, a1, at thatlocus and
that it does not lead to altruistic behavior.
Consider three possibilities: (1) the altruists
dispense their altruism randomly throughout
the population; (2) they dispense it nonrandomly by regarding their degree of genetic
relationshipwith possible recipients;or (3) they
dispense it nonrandomlyby regardingthe altruistictendenciesof possible recipients.
(1) Random dispensationof altruism
There are three possible genotypes: a1a1,
and a2a2' Each allele of the heterozygote
a2al,
will be affectedequally by whatevercosts and
benefitsare associatedwith the altruismof such
individuals (if a2 is dominant)and by whatever
benefitsaccrue to such individuals from the
altruismof others,so theycan be disregarded.
If altruisticacts are being dispensed randomly
throughouta large population, then the typical a1ai individual benefitsby (1/N)lb1, where
bi is the benefitof the ith altruisticact performedby the altruist. The typical a2a2 individual has a net benefitof (l/N)Jbi - (I/N)Icj,
wherec; is the cost to the a2a2 altruistof his jth
altruisticact. Since -(1 /N)Icj is always less
than zero, allele a1 will everywherereplace
allele a2.
(2) Nonrandom dispensationby reference
to kin
This case has been treated in detail by
Hamilton (1964), who concluded that if the
tendencyto dispense altruism to close kin is
great enough, as a function of the disparity
between the average cost and benefit of an
altruistic act, then a2 will replace a1. Tech-
MARCH
RECIPROCAL
1971]
nically,all that is needed for Hamilton's form
of selection to operate is that an individual
with an "altruisticallele" be able to distinguish
between individuals with and without this allele and discriminateaccordingly. No formal
analysishas been attemptedof the possibilities
for selection favoringindividuals who increase
theirchances of receivingaltruisticacts by appearing as if they were close kin of altruists,
although selection has clearly sometimes favored such parasitism (e.g., Drury and Smith,
1968).
(3) Nonrandom dispensation by referenceto
the altruistic tendencies of the recipient
What is required is that the net benefitaccruing to a typicala2a2 altruistexceed that accruing to an alai non-altruist,
or that
(I
/p2)
(lb k -
Cj) > (I /q2)lbm,
wherebk is the benefitto the a2a2
altruist
of the kth altruistic act performed toward
him, where ci is the cost of the jth altruistic
act by the a2a2 altruist,where bmis the benefit
of the mth altruisticact to the alai nonaltruist,and where p is the frequencyin the
population of the a2 allele and q that of the a,
allele. This will tend to occur if Ibm is kept
small (which will simultaneouslyreduce Ici).
And this in turn will tend to occur if an altruistrespondsto a "nonaltruisticact" (that is,
a failureto act altruisticallytowardthe altruist
in a situationin which so doing would cost the
actor less than it would benefitthe recipient)
by curtailingfuturealtruisticacts to the nonaltruist.
Note that the above form of altruismdoes
not depend on all altruistic acts being controlled by the same allele at the same locus.
Each altruistcould be motivatedby a different
allele at a differentlocus. All altruisticalleles
would tend to be favoredas long as, for each
allele, the net average benefitto the homozygous altruistexceeded the averagebenefitto the
homozygousnonaltruist;this would tend to be
true if altruistsrestricttheiraltruismto fellow
altruists,regardlessof what allele motivatesthe
other individual's altruism. The argumentwill
thereforeapply, unlike Hamilton's (1964), to
altruisticacts exchanged between membersof
different
species. It is the exchange that favors
such altriusm,not the fact that the allele in
ALTRUISM
37
question sometimesor oftendirectlybenefitsits
duplicatein anotherorganism.
If an "altruisticsituation" is definedas any
in which one individual can dispense a benefit
to a second greaterthan the cost of the act to
himself,then the chances of selecting for altruisticbehavior, that is, of keeping jcj+?bm
small,are greatest(1) when thereare manysuch
altruisticsituations in the lifetimeof the altruists,(2) when a given altruist repeatedly
interactswith the same small set of individuals,
and (3) when pairs of altruistsare exposed "symmetrically"to altruisticsituations,that is, in
such a way that the two are able to render
roughly equivalent benefitsto each other at
roughly equivalent costs. These three conditions can be elaborated into a set of relevant
biological parameters affectingthe possibility
that reciprocally altruistic behavior will be
selected for.
(1) Length of lifetime.Long lifetimeof individuals of a species maximizesthe chance that
any two individuals will encounter many altruisticsituations,and all other things being
equal one should searchforinstancesof reciprocal altruismin long-livedspecies.
(2) Dispersal rate. Low dispersalrate during
all or a significantportion of the lifetimeof
individuals of a species increases the chance
that an individual will interactrepeatedlywith
the same set of neighbors,and other thingsbeing equal one should search for instances of
reciprocalaltruismin such species. Mayr (1963)
has discussedsome of the factorsthatmay affect
dispersalrates.
(3) Degree of mutual dependence. Interdependence of members of a species (to avoid
predators,for example) will tend to keep individuals near each other and thus increase the
chance theywill encounteraltruisticsituations
together. If the benefit of the mutual dependence is greatestwhen only a small number
of individuals are together,this will greatly
increase the chance that an individual will repeatedly interact with the same small set of
individuals. Individuals in primate troops,for
example, are mutually dependent for protection frompredation,yet the optimal troop size
for foragingis often small (Crook, 1969). Because theyalso meet the other conditionsoutlined here, primatesare almost ideal species in
which to search for reciprocalaltruism. Clean-
38
THE QUARTERLY
REVIEW OF BIOLOGY
[VOLUME46
ing symbiosesprovide an instance of mutual to that of the modern chimpanzeethan to that
dependence between membersof differentspe- of the modern baboon (see, for example, Reycies, and this mutual dependence appears to nolds, 1966).
(6) Aid in combat. No matter how domihave set the stage for the evolution of several
nance-orienteda species is, a dominant indialtruisticbehaviors discussedbelow.
(4) Parental care. A special instance of mu- vidual can usually be aided in aggressiveentual dependence is that found between parents counterswith other individualsby help froma
in speciesthatshow parentalcare. less dominant individual. Hall and DeVore
and offspring
that (1965) have describedthe tendencyfor baboon
The relationshipis usuallyso asymmetrical
fewor no situationsarise in which an offspring alliances to formwhich fightas a unit in agis capable of performing
an altruisticact forthe gressive encounters (and in encounters with
parentsor even for another offspring,
but this predators). Similarly,vervet monkeys in agis not entirelytrue for some species (such as gressive encounters solicit the aid of other,
primates)in which the period of parental care often less dominant, individuals (Struhsaker,
is unusuallylong. Parental care,of course,is to 1967). Aid in combat is then a special case in
be explained by Hamilton's (1964) model, but which relativelysymmetricalrelations are posthere is no reason why selection for reciprocal sible between individuals who differin domialtruismcannot operate betweenclose kin, and nance.
evidence is presentedbelow that such selection
The above discussionis meant only to suggest
has operatedin humans.
the broad conditions that favor the evolu(5) Dominance hierarchy.Linear dominance tion of reciprocal altruism. The most imporhierarchiesconsist by definitionof asymmetri- tant parametersto specifyfor individuals of a
cal relationships;a given individual is domi- species are how many altruisticsituationsoccur
nant over another but not vice versa. Strong and how symmetrical
theyare, and theseare the
dominance hierarchies reduce the extent to most difficultto specifyin advance. Of the
which altruisticsituationsoccur in which the three instancesof reciprocal altruismdiscussed
less dominantindividual is capable of perform- in this paper only one, human altruism,would
ing a benefitfor the more dominantwhich the have been predicted from the above broad
more dominant individual could not simply conditions.
take at will. Baboons (Papio cynocephalus)proThe relationship between two individuals
vide an illustrationof this. Hall and DeVore repeatedly exposed to symmetricalreciprocal
(1965) have described the tendency for meat situations is exactly analogous to what game
caughtby an individual in the troop to end up theoristscall the Prisoner'sDilemma (Luce and
by preemptionin the hands of the most domi- Raiffa, 1957; Rapoport and Chammah, 1965),
nant males. This ability to preempt removes a game that can be characterizedby the payoff
any selectiveadvantage that food-sharing
might matrix
otherwisehave as a reciprocal gesturefor the
A2
C2
most dominantmales, and thereis no evidence
S, T
R, R
in this species of any food-sharingtendencies.
Al
By contrast, Van Lawick-Goodall (1968) has
C1
P, P
T, S
shown that in the less dominance-oriented
chimpanzeesmore dominant individuals often where S < P < R < T and where A1 and A2
do not preempt food caught by the less domi- representthe altruisticchoices possible for the
nant. Instead, they besiege the less dominant two individuals, and C1 and C2, the cheating
individual with "begging gestures,"which re- choices (the firstletter in each box gives the
sult in the handing over of small portions of payofffor the firstindividual,the second letter
the catch. No strongevidence is available that the payoff for the second individual). The
this is part of a reciprocallyaltruisticsystem, othersymbolscan be given the followingmeanbut the absence of a stronglinear dominance ings: R stands for the reward each individual
hierarchyhas clearly facilitatedsuch a possi- gets from an altruistic exchange if neither
bility. It is very likely that early hominid cheats; T stands for the temptationto cheat;
groups had a dominance systemmore similar S stands for the sucker'spayoffthat an altruist
MARCH
1971]
RECIPROCAL
gets when cheated; and P is the punishment
that both individuals get when neither is altruistic (adapted from Rapoport and Chammah, 1965). Iteratedgames played between the
same two individuals mimic real life in that
theypermiteach player to respond to the behavior of the other. Rapoport and Chammah
(1965) and othershave conducted such experiments using human players,and some of their
resultsare reviewedbelow in the discussionof
human altruism.
W. D. Hamilton (pers. commun.) has shown
that the above treatmentof reciprocalaltruism
can be reformulatedconciselyin termsof game
theoryas follows. Assuming two altruistsare
symmetrically
exposed to a series of reciprocal
situations with identical costs and identical
benefits,then after 2n reciprocal situations,
each has been "paid" nR. Were one of the two
a nonaltruistand the second changed to a nonaltruisticpolicy after firstbeing cheated, then
the initial altruistwould be paid S + (n - I)P
(assuming he had the firstopportunityto be
altruistic) and the non-altruistwould receive
T + (n - I)P. The important point here is
that unless T >> R, then even with small n,
nR should exceed T + (n - I)P. If this holds,
the nonaltruistictype,when rare, cannot start
to spread. But there is also a barrier to the
spread of altruismwhen altruistsare rare, for
P > S implies nP > S + (n-1)P.
As n increases,thesetwo total payoffstend to equality,
so the barrierto the spread of altruismis weak
if n is large. The barrierwill be overcome if
the advantages gained by exchanges between
altruists outweigh the initial losses to nonaltruistictypes.
Reciprocal altruismcan also be viewed as a
symbiosis,each partnerhelping the otherwhile
he helps himself.The symbiosishas a time lag,
however;one partnerhelps the otherand must
then wait a period of time beforehe is helped
in turn. The returnbenefitmay come directly,
as in human food-sharing,
the partner directly
returningthe benefitaftera time lag. Or the
returnmay come indirectly,as in warningcalls
in birds (discussed below), where the initial
help to other birds (the warning call) sets up
a causal chain through the ecological system
(the predatorfails to learn useful information)
which redounds aftera time lag to the benefit
of the caller. The timelag is the crucial factor,
39
ALTRUISAM
for it means that only under highlyspecialized
circumstancescan the altruist be reasonably
guaranteed that the causal chain he initiates
with his altruisticact will eventuallyreturnto
him and confer,directlyor indirectly,its benefit. Only under these conditions will the
cheater be selected against and this type of
altruisticbehavior evolve.
Althoughthe preconditionsforthe evolution
of reciprocal altruism are specialized, many
species probably meet them and display this
type of altruism. This paper will limit itself,
however, to three instances. The first,behavior involvedin cleaning symbioses,is chosen
because it permits a clear discriminationbetween this model and that based on kin selection (Hamilton, 1964). The second, warning
calls in birds, has already been elaborately
analyzed in terms of kin selection; it is discussed here to show how the model presented
above leads to a very differentinterpretation
of these familiar behaviors. Finally, human
reciprocal altruism is discussed in detail because it representsthe best documentedcase of
reciprocal altruism known, because there has
apparently been strong selection for a very
complex systemregulating altruisticbehavior,
and because the above model permitsthe functional interpretationof details of the system
that otherwiseremain obscure.
ALTRUISTIC
BEHAVIOR
IN CLEANING
SYMBIOSES
The preconditionsfor the evolution of reciprocal altruism are similar to those for the
operation of kin selection: long lifetime,low
dispersal rate, and mutual dependence, for example, tend to increase the chance that one is
interactingwith one's close kin. This makes it
difficult
to discriminatethe two alternativehypotheses. The case of cleaning symbiosisis
importantto analyze in detail because altruistic
behavior is displayed that cannot be explained
by kin selection,since it is performedby members of one species for the benefitof members
of another. It will be shown instead that the
behavior can be explained by the model presented above. No elaborate explanation is
needed to understand the evolution of the
mutually advantageous cleaning symbiosisitself; it is several additional behaviorsdisplayed
by the host fish to its cleaner that require a
40
THE QUARTERLY REVIEW OF BIOLOGY
special explanation because they meet the
criteriaforaltruisticbehavioroutlined abovethatis, theybenefitthecleanerwhileapparently
beingdetrimentalto thehost.
Feder (1966) and Maynard (1968) have recentlyreviewedthe literatureon cleaning symbioses in the ocean. Briefly,
one organism(e.g.,
the wrasse, Labroides dimidiatus) cleans another organism(e.g., the grouper,Epinephelus
striatus) of ectoparasites(e.g., caligoid copepods), sometimesenteringinto the gill chambersand mouthof the "host" in orderto do so.
Over forty-five
species of fishare known to be
cleaners,as well as six species of shrimp. Innumerablespecies of fishserve as hosts. Stomach analyses of cleaner fish demonstratethat
theyvary greatlyin the extent to which they
depend on theircleaning habits forfood,some
apparentlysubsistingnearlyentirelyon a diet
of ectoparasites.Likewise,stomachanalyses of
host fishreveal that cleaners differin the rate
at which theyend up in the stomachsof their
hosts, some being apparently almost entirely
immuneto such a fate. It is a strikingfactthat
thereseems to be a strongcorrelationbetween
degree of dependence on the cleaning way of
life and immunityto predation by hosts.
Cleaning habits have apparentlyevolved independentlymanytimes(at least threetimesin
shrimps alone), yet some remarkable convergence has taken place. Cleaners, whether
shrimp or fish,are distinctivelycolored and
behave in distinctiveways (for example, the
wrasse,L. dimidiatus,swimsup to its host with
a curious dipping and rising motion that reminds one of the way a finchflies). These distinctivefeaturesseem to serve the functionof
attractingfishto be cleaned and of inhibiting
any tendencyin themto feed on theircleaners.
There has apparentlybeen strongselection to
avoid eating one's cleaner. This can be illustrated by several observations. Hediger
(1968) raised a grouper (Epinephelus) from
infancyalone in a small tank for six years,by
which time the fish was almost four feet in
length and accustomed to snapping up anything dropped into its tank. Hediger then
dropped a small live cleaner (L. dimidiatus)
into the grouper'stank. The groupernot onily
failed to snap up the cleaner but opened its
mouthand permittedthe cleanerfreeentryand
exit.
[VOLUME
46
Soonwe watchedoursecondsurprise:thegrouper
made a movementwhichin the precedingsix
yearswe had neverseenhimmake:he spreadthe
rightgill-covering
so wide that the individual
gill-plateswere separatedfromeach other at
greatdistances,wide enoughto let the cleaner
through(translatedfromHediger,1968 p. 93).
When Hediger added two additional L. dimidiatus to the tank, all three cleaned the
grouper with the result that within several
days the grouperappeared restlessand nervous,
searchedout places in the tank he had formerly
avoided, and shook himselfoften (as a signal
thathe did not wish to be cleaned any longer).
Apparently three cleaners working over him
constantlywas too much for him, yet he still
failed to eat any of them. When Hediger removed two of the cleaners, the grouper returned to normal. There is no indication the
grouperever possessedany edible ectoparasites,
and almosttwo yearslater (in December, 1968)
the same cleaner continued to "clean" the
grouper (pers. observ.) although the cleaner
was, in fact,fed separatelyby its zoo-keepers.
Eibl-Eibesfeldt(1959) has describedthe morphology and behavior of two species (e.g., Aspidontutstaeniatus) that mimic cleaners (e.g.,
L. dimidiatus) and that rely on the passive
behavior of fishwhich suppose theyare about
to be cleaned to dart in and bite offa chunk
of their fins. I cite the evolution of these
mimics, which resemble their models in appearance and initial swimmingbehavior, as
evidence of strongselectionfor hosts with no
intentionof harmingtheir cleaners.
Of especial interestis evidence that therehas
been strongselection not to eat one's cleaner
even afterthe cleaning is over. Eibl-Eibesfeldt
(1955) has made some strikingobservationson
the goby,Elacitinus oceanops:
I neversawv
a groupersnap up a fishafterit had
cleaned it. On the contrary,
it annoounced
its
impending
departure
by twodefinite
signalmovements. Firstit closed its mouthvigorously,
aland immediately
thoughnot completely,
opened
it wide again. Upon this intentionmovement,
all the gobiesleftthe mouthcavity.Then the
groupershookits bodylaterallya fewtines,and
all the cleaniers
returnedto theircoral. If onie
a grouperit never neglectedthese
frightened
movements(translatedfrom Eiblforewarning
Eibesfeldt,
1955,p. 208).
MARCH
1971]
RECIPROCAL
Randall has made similar observationson a
moray eel (Gymnothoraxjaponicus) that signalled with a "sharp lateral jerk of the eel's
head," afterwhich "the wrasse fairlyflew out
of the mouth, and the awesome jaws snapped
shut" (Randall, 1958, 1962). Likewise, Hediger's Kasper Hauser grouper shook its body
when it had enough of being cleaned.
Why does a large fishnot signal the end to a
cleaning episode by swallowing the cleaner?
Natural selection would seem to favor the
double benefitof a good cleaning followed by
a meal of the cleaner. Selection also operates,
of course,on the cleaner and presumablyfavors
mechanismsto avoid being eaten. The distinctive behavior and appearance of cleaners has
been cited as evidence of such selection. One
can also cite the distinctivebehaviorof the fish
being cleaned. Feder (1966) has pointed out
that hostsapproachinga cleaner react by "stopping or slowing down, allowing themselvesto
assume awkwardpositions,seeminglyin a hypnotic state." Fishes sometimesalter their color
dramaticallybefore and while being cleaned,
and Feder (1966) has summarizedinstancesof
this. These forms of behavior suggest that
natural selection has operated on cleaners to
avoid attemptingto clean fish without these
behaviors,presumablyto avoid wasting energy
and to minimize the dangers of being eaten.
(Alternatively,the behaviors, including color
change, may aid the cleaners in findingectoparasites. This is certainlypossible but not,
I believe, adequate to explain the phenomenon
completely. See, for example, Randall, 1962.)
Once the fishto be cleaned takes the proper
stance,however,the cleaner goes to work with
no apparent concernforits safety: it makes no
effortto avoid the dangerous mouth and may
even swim inside,which as we have seen, seems
particularlyfoolhardy,since fishbeing cleaned
may suddenly need to depart. The apparent
unconcernof the cleaner suggeststhat natural
selection acting on the fishbeing cleaned does
not, in fact, favor eating one's cleaner. No
speculation has been advanced as to why this
may be so, although some speculation has appeared about the mechanismsinvolved. Feder
advances two possibilities, that of EiblEibesfeldt (1955) that fishcome to be cleaned
only aftertheirappetite has been satisfied,and
one of his own, that the irritationof ecto-
ALTRUISM
41
parasites may be sufficientto inhibit hunger.
Both possibilitiesare contradictedby Hediger's
observation,cited above, and seem unlikelyon
functionalgroundsas well.
A fishto be cleaned seems to performseveral
"altruistic" acts. It desists from eating the
cleaner even when it easily could do so and
when it must go to special pains (sometimes
at danger to itself)to avoid doing so. Furthermore,it may performtwo additional behaviors
which seem of no direct benefitto itself (and
which consume energyand take time), namely,
it signals its cleaner that it is about to depart
even when the fishis not in its mouth,and it
may chase offpossible dangers to the cleaner:
While diving with me in the VirginIslands,
Robert Schroederwatched a Spanish hogfish
groominga bar jack in its bronze color state.
When a secondjack arrivedin the pale color
droveit away.
phase,the firstjack immediately
But later when anotherjack intrudedon the
sceneand changedits pale color to dark bronze
it was not chased. The bronzecolorwould seem
to mean "no harm intended;I need service"
(Randall,1962p. 44).
The behavior of the host fishis interpreted
here to have resulted from natural selection
and to be, in fact,beneficialto the hostbecause
the cleaner is worthmore to it alive than dead.
This is because the fishthat is cleaned "plans"
to returnat later dates formore cleanings,and
it will be benefitedby being able to deal with
the same individual. If it eats the cleaner, it
findinga second when it
may have difficulty
needs to be cleaned again. It may lose valuable
energy and be exposed to unnecessarypredation in the search for a new cleaner. And it
may in the end be "turned down" by a new
cleaner or servicedvery poorly. In short,the
host is abundantly repaid for the cost of its
altruism.
To support the hypothesisthat the host is
repaid its initial altruism,several pieces of evidence mustbe presented: that hostssufferfrom
ectoparasites;that findinga new cleaner may
or dangerous; that if one does not
be difficult
eat one's cleaner,the same cleanercan be found
and used a second time (e.g., that cleaners are
site-specific);that cleaners live long enough to
be used repeatedly by the same host; and if
possible, that individual hostsdo, in fact,reuse
the same cleaner.
42
THE QUARTERLY
REVIEW OF BIOLOGY
[VOLUME
46
Cleaning fishesand cleaning shrimpshave regular
(1) The cost of ectoparasites. It seems alto which fishes wanting to be cleaned
stations
most axiomatic that the evolution of cleaners
can come (p. 367).
entirely dependent on ectoparasitesfor food
implies the selective disadvantage for the Limbaugh, Pederson, and Chase (1961) have
cleaned of being ectoparasite-ridden.What is reviewed available data on the six species of
perhaps surprisingis the effectthat removing cleaner shrimps, and say:
all cleaners froma coral reef has on the local
The known cleaner shrimpsmay convenientlybe
"hosts" (Limbaugh, 1961). A, '<eder (1966) said
in his review:
Within a few days the numberof fisheswas
drasticallyreduced. Within two weeks almost
fisheshad disappeared,and
all exceptterritorial
manyofthesehad developedwhitefuzzyblotches,
ulceratedsores,and frayedfins(p. 366).
swellings,
Clearly, once a fish'sprimaryway of dealing
with ectoparasites is by being cleaned, it is
quickly vulnerable to the absence of cleaners.
and danger of finding a
(2) The difficulty
cleaner. There are naturallyveryfew data on
the difficulty
or dangerof findinga new cleaner.
This is partiallybecause, as shown below, fish
tend repeatedlyto returnto familiar cleaners.
The only observationof fishbeing disappointed
in their search for cleaners comes from EiblEibesfeldt (1955): "If the cleaners fail to appear over one coral in about half a minute,the
large fishes swim to another coral and wait
therea while" (translatedfromp. 210). It may
be that fish have several alternative cleaning
stationsto go to, since any particularcleaning
station may be occupied or unattended at a
givenmoment. So manyfishtend to be cleaned
at coral reefs (Limbaugh, 1961, observed a
cleaner service300 fishin a 6-hourperiod), that
predatorsprobablyfrequentcoral reefsin search
of fishbeing cleaned. Limbaugh (1961) suggested
that good human fishingsites are found near
cleaning stations. One finalreason whycoming
to be cleaned maybe dangerousis thatsome fish
must leave their element to do so (Randall,
1962):
Most impressive were the visits of moray eels,
which do not ordinarilyleave their holes in the
reef during daylight hours, and of the big jacks
which swam up from deeper water to the reef's
edge to be "serviced" before going on their way
(p. 43).
(3) Site specificityof cleaners. Feder (1966)
has reviewed the strikingevidence for the site
specificity
of cleaners and concludes:
divided into two groups on the basis of behavior,
habitat and color. The five species comprising
one group are usually solitary or paired....
All five species are territorial and remain for
weeks and, in some cases, months or possibly
years within a meter or less of the same spot.
They are omnivorous to a slight extent but seem
to be highly dependent upon their hosts for
food. This group is tropical, and the individuals
are brightlymarked. They display themselvesto
their hosts in a conspicuous manner. They probably rarely serve as prey for fishes. A single
species, Hippolysmata californica, comprises the
second group. . . . This species is a gregarious,
wandering, omnivorous animal . . . and is not
highly dependent upon its host for survival. So
far as is known, it does not display itself to
attractfishes(p. 238).
It is H. californica that is occasionally found in
the stomachs of at least one of its hosts. The
striking correlation of territoriality and solitariness with cleaning habits is what theory
would predict. The same correlation can be
found in cleaner fish. Labroides, with four
species, is the genus most completely dependent
on cleaning habits. No Labroides has ever been
found in the stomach of a host fish. All species
are highly site-specific and tend to be solitary.
Randall
(1958) reports that an individual
L. dimidiatus may sometimes swim as much as
60 feet from its cleaning station, servicing fish
on the way. But he notes,
This was especially true in an area where the
highly territorialdamsel fish Pomacentris nigricans (Lepede) was common. As one damsel fish
was being tended, another nearby would assume
a stationarypose with finserect and the Labroides
would move oni to the latter with little hesitation (p. 333).
Clearly, what matters for the evolution of reciprocal altruism is that the same two individuals interact repeatedly. This will be facilitated by the site specificityof either individual.
Of temperate water cleaners, the species most
MARCH
1971]
RECIPROCAL
specialized to cleaning is also apparently the
most solitary(Hobson, 1969).
(4) Lifespan of cleaners. No good data exist
on how long cleaners live, but several observations on both fishand shrimpsuggestthat they
easily live long enough for effectiveselection
against cheaters. Randall (1958) repeatedly
checkedseveralledges and found that different
feedingstationswere occupied for"long periods
of time," apparentlyby the same individuals.
One such feeding station supported two individuals for over threeyears. Of one species of
cleaner shrimp,Stenopus hispidus, Limbaugh,
Pederson, and Chase (1961) said that pairs of
individuals probably remain months, possibly
years,within an area of a square meter.
(5) Hosts using the same cleaner repeatedly.
There is surprisinglygood evidence that hosts
reuse the same cleaner repeatedly.Feder (1966)
summarizesthe evidence:
43
ALTRUISM
of cleaning organismsperformseveral kinds of
altruisticbehavior, including not eating their
cleaner after a cleaning, which can be explained on the basis of the above model. A
review of the relevant evidence suggeststhat
the cleaner organismsand theirhosts meet the
preconditionsfor the evolution of reciprocally
altruisticbehavior. The host's altruismis to be
explained as benefitinghim because of the
advantage of being able quicklyand repeatedly
to returnto the same cleaner.
WARNING
CALLS
IN BIRDS
Marler (1955, 1957) has presented evidence
that warningcalls in birds tend to have characteristicsthat limit the informationa predator
gets fromthe call. In particular,the call characteristicsdo not allow the predator easily to
determinethe location of the call-giver.Thus,
it seems that givinga warningcall must result,
at least occasionally,in the otherwiseunnecesManyfishesspendas muchtimegettingcleaned sarydeath of the call-giver,eitherat the hands
as theydo foragingforfood. Some fishesreturn of the predatorthat inspired the call or at the
again and again to the same station,and showa
unaware of
hands of a second predatorformerly
definitetime pattern in their daily arrival.
exact
location.
the
caller's
or
presence
Otherspass fromstationto stationand return
manytimesduringthe day; this is particularly Given the presumed selection against callgiving, Williams (1966) has reviewed various
trueof an injuredor infectedfish(p. 368).
models to explain selection for warning cries:
Limbaugh, Pederson, and Chase (1961) have
(1) Warning calls are functionalduring the
presentedevidence that in at least one species breeding season in birds in that they protect
of cleaner shrimp (Stenopus scutellus), the one's mate and offspring.They have no funcshrimp may reservice the same individuals: tion outside the breeding season, but they are
One pair was observedin the same football-sized not deleted then because "in practice it is not
coral boulder fromMay throughAugust 1956. worthburdening the germ plasm with the inDuringthatperiod,we changedthe positionand formationnecessaryto realize such an adjustorientation
of the boulderseveraltimeswithina ment" (Williams, 1966,p. 206).
radius of approximately
seven meterswithout
(2) Warning calls are selected for by the
disturbingthe shrimp.Visitingfisheswere mo- mechanismof group selection(Wynne-Edwards,
mentarilydisturbedby the changes,but they 1962).
soon relocatedthe shrimps(p. 254).
(3) Warning calls are functionaloutside the
Randall (1958) has repeatedly observed fish breedingseason because thereis usually a good
swimmingfromout of sight directlyto clean- chance that a reasonably close kin is near
ing stations,behavior suggestingto him that enough to be helped sufficiently(Hamilton,
theyhad prior acquaintance with the stations. 1964; Maynard Smith, 1964). Maynard Smith
During two months of observationsat several (1965) has analyzed in great detail how closely
feeding stations,Eibl-Eibesfeldt(1955) became related the benefitedkin mustbe, at what benepersonally familiar with several individual fit to him the call must be, and at what cost
groupers(Epinephelus striatus)and repeatedly to the caller, in order for selection to favor
observed them seeking out and being cleaned call-giving.
at the same feedingstations,presumablyby the
The firstis an explanation of last resort.
While it must sometimesapply in evolutionary
same cleaners.
In summary,it seemsfairto say that the hosts arguments,it should probablyonly be invoked
44
THE QUARTERLY
REVIEW OF BIOLOGY
when no otherexplanationseemsplausible. The
second is not consistentwith the known workings of natural selection. The thirdis feasible
and may explain the warning calls in some
species and perhaps even in many. But it does
depend on the somewhatregular nearby presence of closelyrelated organisms,a matterthat
may oftenbe the case but thathas been demonstratedonly as a possibilityin a fewspeciesand
that seems very unlikely in some. A fourth
explanation is suggestedby the above model:
(4) Warning calls are selected for because
theyaid the bird givingthe call. It is disadvantageous for a bird to have a predator eat a
nearby conspecificbecause the predator may
then be more likelyto eat him. This may happen because the predator will
(i) be sustained by the meal,
(ii) be more likelyto forma specificsearch
image of the prey species,
(iii) be more likely to learn the habits of
the preyspecies and perfecthis predatorytechniqueson it,
(iv) be more likely to frequentthe area in
which the birds live, or
(v) be more likelyto learn usefulinformation about the area in which the birds
live.
In short,in one way or another,givinga warning call tendsto preventpredatorsfromspecializing on the caller's species and locality.
There is abundant evidence for the importance of learning in the lives of predatoryvertebrates (see, for example, Tinbergen, 1960;
Leyhausen, 1965; Brower and Brower, 1965).
Rudebeck (1950, 1951) has presentedimportant
observationson the tendencyof avian predators
to specialize individually on prey types and
hunting techniques. Owen (1963) and others
have presented evidence that species of snails
and insects may evolve polymorphismsas a
protectionagainst the tendencyof their avian
predatorsto learn their appearance. Similarly,
Kuyton (1962; cited in Wickler, 1968) has describedthe adaptation of a moththatminimizes
the chance of its predatorsforminga specific
search image. Southern (1954), Murie (1944),
and numerous others have documented the
tendencyof predators to specialize on certain
localities within their range. Finally, Blest
[VOLUME
46
(1963) has presentedevidence thatkin selection
in some cryptic saturnid moths has favored
rapid, post-reproductivedeath to minimize
predation on the young. Blest's evidence thus
providesan instanceof a predatorgaining useful informationthroughthe act of predation.
It does not matterthat in giving a warning
call the caller is helping its non-callingneighbors more than it is helping itself.What counts
is that it outcompetesconspecificsfrom areas
in which no one is giving warning calls. The
non-calling neighbors of the caller (or their
offspring)will soon find themselvesin an area
withoutany caller and will be selected against
relative to birds in an area with callers. The
caller, by definition,is always in an area with
at least one caller. If we assume that two callers
are preferableto one, and so on, then selection
will favorthe spread of the warning-callgenes.
Note that this model depends on the concept
of open groups, whereas "group selection"
(Wynne-Edwards,1962) depends partly on the
concept of closed groups.
It mightbe supposed that one could explain
bird calls more directlyas altruisticbehavior
that will be repaid when the other birds reciprocate, but there are numerous objections
to this. It is difficult
to visualizehow one would
discover and discriminateagainst the cheater,
and there is certainlyno evidence that birds
refrainfromgiving calls because neighborsare
not reciprocating.Furthermore,if the relevant
bird groupingsare very fluid,with much emigration and immigration,as they often are,
then cheating would seem to be favored and
no selectionagainstit possible. Instead, according to the model above, it is the mere factthat
the neighborsurvivesthat repays the call-giver
his altruism.
It is almost impossibleto gather the sort of
evidence that would discriminatebetween this
explanation and that of Hamilton (1964). It
is difficultto imagine how one would estimate
the immediatecost of giving a warningcall or
its benefitto those within earshot,and precise
data on the geneticrelationshipsof bird groupings throughoutthe year are not only lacking
but would be most difficultto gather. Several
lines of evidence suggest,however,that Hamilton's (1964) explanation should be assumed
with caution:
MARCH
RECIPROCAL
1971]
(1) There exist no data showing a decrease
in warning tendencies with decrease in
the genetic relationshipof those within
earshot. Indeed, a striking feature of
warningcalls is thattheyare given in and
out of the breeding season, both before
and after migrationor dispersal.
(2) There do exist data suggestingthat close
kin in a number of species migrate or
disperse great distances from each other
(Ashmole, 1962; Perdeck, 1958; Berndt
and Sternberg,1968; Dhont and Huble,
1968).
(3) One can advance the theoretical argument that kin selection under some circumstancesshould favor kin dispersal in
order to avoid competition (Hamilton,
1964, 1969). This would lead one to
expect fewercloselyrelated kin near any
given bird, outside the breeding season.
The argumentsadvanced in this section may
also apply,of course,to speciesotherthan birds.
HUMAN
RECIPROCAL
ALTRUISM
Reciprocal altruism in the human species
takes place in a numberof contextsand in all
known cultures (see, for example, Gouldner,
1960). Any complete list of human altruism
would contain the followingtypesof altruistic
behavior:
(1) helping in timesof danger (e.g. accidents,
predation, intraspecificaggression;
(2) sharing food;
(3) helping the sick, the wounded, or the
veryyoungand old;
(4) sharing implements;and
(5) sharingknowledge.
All these forms of behavior often meet the
criterionof small cost to the giver and great
benefitto the taker.
During the Pleistocene,and probablybefore,
a hominid species would have met the preconditionsforthe evolutionof reciprocalaltruism:
long lifespan; low dispersal rate; life in small,
mutuallydependent,stable, social groups (Lee
and DeVore, 1968; Campbell, 1966); and a long
period of parental care. It is very likely that
dominance relations were of the relaxed, less
linear formcharacteristic
of the livingchimpan-
ALTRUISM
45
zee (Van Lawick-Goodall, 1968) and not of the
more rigidlylinear form characteristicof the
baboon (Hall and DeVore, 1965). Aid in intraspecificcombat,particularlyby kin, almost certainlyreduced the stabilityand linearityof the
dominance order in early humans. Lee (1969)
has shown that in almost all Bushman fights
which are initially between two individuals,
othershave joined in. Mortality,for example,
often strikes the secondaries rather than the
principals. Tool use has also probably had an
equalizing effecton human dominance relations, and the Bushmen have a saying that
illustratesthis nicely. As a dispute reaches the
stage where deadly weapons may be employed,
an individualwill oftendeclare: "We are none
of us big, and otherssmall; we are all men and
we can fight;I'm going to get myarrows,"(Lee,
1969). It is interestingthatVan Lawick-Goodall
(1968) has recordedan instanceof strongdominance reversalin chimpanzeesas a functionof
tool use. An individual moved from low in
dominance to the top of the dominance hierarchywhen he discoveredthe intimidatingeffects of throwinga metal tin around. It is
likely that a diversityof talentsis usually present in a band of hunter-gatherers
such that the
best makerof a certaintypeof tool is not often
the best maker of a differentsort or the best
user of the tool. This contributesto the symmetaryof relationships,since altruisticacts can
be traded with referenceto the special talents
of the individuals involved.
To analyze the details of the human reciprocal-altruisticsystem,several distinctionsare important and are discussedhere.
(1) Kin selection. The human species also
met the preconditionsfor the operation of kin
selection. Early hominid hunter-gatherer
bands
almost certainly(like today's hunter-gatherers)
consistedof many close kin, and kin selection
mustoftenhave operated to favorthe evolution
of some types of altruisticbehavior (Haldane,
1955; Hamilton, 1964, 1969). In general, in
attemptingto discriminatebetween the effects
of kin selection and what might be called reciprocal-altruistic
selection,one can analyze the
formof the altruisticbehaviorsthemselves.For
example,the existenceof discriminationagainst
non-reciprocalindividuals cannot be explained
on the basis of kin selection,in which the advantage accruingto close kin is what makes the
46
THE QUARTERLY
REVIEW OF BIOLOGY
[VOLUME
46
altruisticbehavior selectivelyadvantageous,not should show interestingdifferences,but the
its chance of being reciprocated.The strongest analysis is complicatedby the possibilityof reargumentfor the operation of reciprocal-altru- ciprocityto the kin of a deceased altruist(see
isticselectionin man is the psychologicalsystem Multi-partyinteractionsbelow).
(4) Gross and subtle cheating. Two formsof
controlling some forms of human altruism.
cheatingcan be distinguished,here denoted as
Details of thissystemare reviewedbelow.
(2) Reciprocal altruismamong close kin. If grossand subtle. In gross cheatingthe cheater
both forms of selection have operated, one failsto reciprocateat all, and the altruistsuffers
the costsof whateveraltruismhe has dispensed
would expect some interestinginteractions.One
might expect, for example, a lowered demand without any compensating benefits. More
forreciprocityfromkin than fromnonkin,and broadly, gross cheating may be defined as rethereis evidence to supportthis (e.g., Marshall, ciprocatingso little, if at all, that the altruist
1961; Balikci, 1964). The demand that kin receivesless benefitfromthe grosscheaterthan
show some reciprocity(e.g., Marshall, 1961; the cost of the altruist'sacts of altruismto the
Balikci, 1964) suggests,however,thatreciprocal- cheater. That is, Icai>
Nbaj where cai is the
altruisticselection has acted even on relations cost of the ith altruisticact performedby the
between close kin. Although interactionsbe- altruistand where
bai is the benefitto the altween the two formsof selectionhave probably truistof the jth altruisticact performedby the
been importantin human evolution,thispaper gross cheater; altruisticsituationsare assumed
will limit itselfto a preliminarydescriptionof to have occurred symmetrically.Clearly, selecthe human reciprocallyaltruisticsystem,a sys- tion will stronglyfavor prompt discrimination
tem whose attributesare seen to result only against the gross cheater. Subtle cheating,by
selection.
from reciprocal-altruistic
contrast,involves reciprocating,but always at(3) Age-dependentchanges. Cost and benefit temptingto give less than one was given, or
were defined above without referenceto the more precisely,to give less than the partner
ages, and hence reproductive values (Fisher, would give if the situation were reversed. In
1958), of the individuals involved in an altru- thissituation,the altruiststill benefitsfromthe
istic exchange. Since the reproductivevalue of relationshipbut not as much as he would if
a sexually mature organismdeclines with age, the relationship were completely equitable.
the benefitto him of a typicalaltruisticact also The subtlecheaterbenefitsmore than he would
decreases,as does the cost to him of a typical if the relationship were equitable. In other
act he performs.If the intervalseparatingthe words,
two acts in an altruisticexchange is shortrelaI
(bqi - cqj) > t (bqi - Cai) > i. (ba - cai)
tive to the lifespans of the individuals, then
the erroris slight. For longerintervals,in order where the ith altruisticact performedby the
to be repaid precisely,the initial altruistmust altruisthas a cost to him of cal and a benefit
receive more in return than he himselfgave. to the subtle cheaterof bqi and where the jth
It would be interestingto see whetherhumans altruisticact performedby the subtle cheater
in fact routinelyexpect "interest"to be added has a cost to him of cq, and a benefitto the
to a long overdue altruisticdebt, interestcom- altruist of baj. Because human altruism may
mensuratewith the interveningdecline in re- span huge periods of time,a lifetimeeven, and
productivevalue. In humansreproductivevalue because thousandsof exchangesmay take place,
declines most steeply shortlyafter sexual ma- involving many different"goods" and with
turityis reached (Hamilton, 1966), and one many differentcost/benefit
ratios,the problem
would predict the interest rate on altruistic of computingthe relevant totals,detectingimdebts to be highestthen. Selection mightalso balances, and deciding whether they are due
favorkeeping the intervalbetween act and re- to chance or to small-scalecheating is an exone. Even then,the altruistis
ciprocation short,but this should also be fa- tremelydifficult
vored to protectagainst complete non-recipro- in an awkwardposition,symbolizedby the folk
cation.W. D. Hamilton (pers.commun.)has sug- saying,"half a loaf is betterthan none," for if
gestedthat a detailed alialysisof age-dependent attempts to make the relationship equitable
changesin kin altruismand reciprocalaltruism lead to the ruptureof the relationship,the al-
MARCH
1971]
RECIPROCAL
truist,assumingother thingsto be equal, will
sufferthe loss of the substandard altruism
of the subtle cheater. It is the subtletyof the
discriminationnecessaryto detect this formof
cheatingand the awkwardsituationthat ensues
thatpermitsome subtlecheatingto be adaptive.
This sets up a dynamic tension in the system
that has importantrepercussions,as discussed
below.
(5) Number of reciprocalrelationships.It has
so far been assumed that it is to the advantage
of each individual to formthe maximumnumber of reciprocal relationships and that the
a decrease in fitnessupon the
individual suffers
rupture of any relationshipin which the cost
to him of acts dispensed to the partneris less
than the benefitof acts dispensed toward him
by the partner. But it is possible that relationships are partly exclusive, in the sense that
expanding the number of reciprocalexchanges
with one of the partnersmay necessarilydecrease the number of exchanges with another.
For example, if a group of organismswere to
split into subgroupsfor much of the day (such
as breakingup into huntingpairs), then altruistic exchanges will be more likely between
membersof each subgroup than between members of differentsubgroups. In that sense, relationships may be partly exclusive, membership in a given subgroup necessarilydecreasing
exchanges with others in the group. The importance of this factor is that it adds further
complexityto the problem of dealing with the
cheater and it increases competitionwithin a
group to be membersof a favorablesubgroup.
An individual in a subgroup who feels that
another member is subtly cheating on their
relationship has the option of attemptingto
restorethe relationshipto a completelyreciprocal one or of attemptingto join another subgroup, therebydecreasing to a minimum the
possible exchanges between himself and the
subtle cheater and replacing these with exchangesbetweena new partneror partners. In
short,he can switchfriends. There is evidence
in hunter-gatherers
that much movement of
individuals from one band to another occurs
in response to such social factorsas have just
been outlined (Lee and DeVore, 1968).
(6) Indirect benefitsor reciprocal altruism?
Given mutual dependence in a group it is possible to argue that the benefits(non-altruistic)
47
ALTRUISM
of this mutual dependence are a positive function of group size and that altruisticbehaviors
may be selected for because they permit additional individuals to surviveand therebyconfer additional indirect (non-altruistic)benefits.
Such an argumentcan only be advanced seriously for slowlyreproducingspecies with little
dispersal. Saving an individual's life in a
hunter-gatherer
group,forexample,maypermit
non-altruisticactions such as cooperativehunting to continue with more individuals. But if
there is an optimum group size, one would
expect adaptations to stay near that size, with
individualsjoining groupswhen the groups are
below this size, and groups splittingup when
they are above this size. One would only be
selected to keep an individual alive when the
group is below optimum and not when the
group is above optimum. Although an abundant literature on hunter-gatherers
(and also
nonhuman primates)suggeststhat adaptations
exist to regulate group size near an optimum,
there is no evidence that altruisticgesturesare
curtailedwhen groups are above the optimum
in size. Instead, the benefitsof human altruism
are to be seen as comingdirectlyfromreciprocity- not indirectly through non-altruistic
group benefits. This distinctionis important
because social scientistsand philosophershave
tended to deal with human altruismin terms
of the benefitsof living in a group, without
between non-altruisticbenefits
differentiating
and reciprocal benefits (e.g., Rousseau, 1954;
Baier, 1958).
THE
PSYCHOLOGICAL
SYSTEM
RECIPROCAL
UNDERLYING
HUMAN
ALTRUISM
Anthropologistshave recognized the importance of reciprocityin human behavior, but
when theyhave ascribed functionsto such behavior they have done so in terms of group
benefits,reciprocitycementinggroup relations
and encouraginggroup survival. The individual sacrificesso that the group may benefit.
Recently psychologistshave studied altruistic
behavior in order to show what factorsinduce
or inhibit such behavior. No attempthas been
made to show what functionsuch behaviormay
serve,nor to describe and interrelatethe components of the psychologicalsystemaffecting
altruisticbehavior. The purpose of thissection
48
THE QUARTERLY
REVIEW OF BIOLOGY
is to show that the above model for the natural
selectionof reciprocallyaltruisticbehavior can
readilyexplain the functionof human altruistic
behavior and the details of the psychological
systemunderlyingsuch behavior. The psychological data can be organized into functional
categories,and it can be shownthat the components of the systemcomplementeach other in
regulatingthe expressionof altruisticand cheating impulsesto the selectiveadvantage of individuals. No concept of group advantage is
necessaryto explain the function of human
altruisticbehavior.
There is no direct evidence regarding the
degree of reciprocal altruismpracticed during
human evolutionnor its geneticbasis today,but
given the universaland nearlydaily practiceof
reciprocal altruismamong humans today, it is
reasonable to assume thatit has been an important factorin recenthuman evolution and that
the underlyingemotional dispositionsaffecting
altruisticbehaviorhave importantgenetic components. To assume as much allows a number
of predictions.
(1) A complex, regulatingsystem. The human altruisticsystemis a sensitive,unstable
one. Often it will pay to cheat: namely,when
the partnerwill not findout, when he will not
discontinue his altruismeven if he does find
out, or when he is unlikely to survive long
enough to reciprocate adequately. And the
perception of subtle cheating may be very
difficult.Given this unstable characterof the
system,where a degree of cheatingis adaptive,
natural selection will rapidly favor a complex
psychologicalsystemin each individual regulating both his own altruistic and cheating
tendenciesand his responsesto thesetendencies
in others. As selection favorssubtler formsof
cheating, it will favor more acute abilities to
detectcheating. The systemthat resultsshould
simultaneouslyallow the individual to reap
the benefitsof altruisticexchanges,to protect
himselffromgross and subtle formsof cheating, and to practicethoseformsof cheatingthat
local conditions make adaptive. Individuals
will differnot in being altruists or cheaters
but in the degree of altruismtheyshow and in
the conditionsunder which theywill cheat.
The best evidencesupportingtheseassertions
can be found in Kreb's (1970) review of the
relevant psychologicalliterature. Although he
[VOLUME
46
much of the material
organizes it differently,
supportingthe assertionsbelow is taken from
his paper. All referencesto Krebs below are to
this review. Also, Hartshorneand May (19281930) have shownthatchildrenin experimental
situationsdo not divide bimodallyinto altruists
and "cheaters" but are distributednormally;
almostall the childrencheated,but theydiffered
in how much and under what circumstances.
("Cheating" was defined in their work in a
but analogous way).
slightlydifferent
(2) Friendshipand the emotionsof likingand
disliking. The tendency to like others, not
necessarilyclosely related, to form friendships
and to act altruisticallytoward friendsand toward those one likes will be selected for as the
immediateemotional rewardsmotivatingaltruistic behavior and the formationof altruistic
partnerships.(Selection may also favorhelping
strangersor disliked individualswhen theyare
in particularlydire circumstances). Selection
will favora systemwherebythesetendenciesare
sensitive to such parameters as the altruistic
tendencies of the liked individual. In other
words,selectionwill favorliking those who are
themselvesaltruistic.
Sawyer(1966) has shownthatall groupsin all
experimental situations tested showed more
altruisticbehavior toward friendsthan toward
neutral indivduals. Likewise,Friedrichs (1960)
has shown that attractivenessas a friend was
most highly correlated among undergraduates
with altruistic behavior. Krebs has reviewed
other studies that suggestthat the relationship
betweenaltruismand likingis a two-waystreet:
one is more altruistictowardthoseone likes and
one tends to like those who are most altruistic
(e.g., Berkowitz and Friedman, 1967; Lerner
and Lichtman, 1968).
Others (Darwin, 1871; Williams, 1966; and
Hamilton, 1969) have recognized the role
friendshipmightplay in engenderingaltruistic
behavior, but all have viewed friendship(and
intelligence)as prerequisitesfor the appearance
of such altruism. Williams (1966), who cites
Darwin (1871) on the matter,speaks of this behavior as evolving
in aniimalsthat live in stablesocial groupsand
and othermentalqualities
have the intelligence
necessaryto forrna systemof personalfriendships and animositiesthat transcendthe limits
of familyrelationships
(p. 93).
MARCH
19711
RECIPROCAL
This emphasison friendshipand intelligenceas
prerequisitesleads Williams to limit his search
for altruismto the Mammalia and to a "minorityof thisgroup." But accordingto the model
presented above, emotions of friendship(and
hatred) are not prerequisites for reciprocal
altruism but may evolve after a system of
mutual altruism has appeared, as important
waysof regulatingthe system.
(4) Moralistic aggression. Once strong positiveemotionshave evolved to motivatealtruistic
behavior,the altruistis in a vulnerableposition
because cheaterswill be selectedto take advantage of the altruist'spositive emotions. This
in turn sets up a selection pressurefor a protective mechanism. Moralistic aggression and
indignationin humanswas selectedforin order
(a) to counteractthe tendencyof the altruist, in the absence of any reciprocity,to continue to performaltruisticacts for his own
emotionalrewards;
(b) to educate the unreciprocatingindividual by frighteninghim with immediate
harm or with the futureharm of no more
aid; and
(c) in extreme cases, perhaps, to select
directlyagainst the unreciprocatingindividual by injuring,killing,or exiling him.
Much of human aggressionhas moral overtones. Injustice, unfairness,and lack of reciprocityoften motivate human aggressionand
indignation. Lee (1969) has shown that verbal
disputes in Bushmen usually revolve around
problemsof gift-giving,
stinginess,and laziness.
DeVore (pers. commun.) reports that a great
deal of aggressionin hunter-gatherers
revolves
around real or imagined injustices-inequities,
for example, in food-sharing(see, for example,
Thomas, 1958; Balikci, 1964; Marshall, 1961).
A common featureof this aggressionis that it
oftenseemsout of all proportionto the offenses
committed. Friends are even killed over apparently trivial disputes. But since small inequities repeated many times over a lifetime
may exact a heavy toll in relative fitness,selection may favor a strong show of aggression
when the cheating tendency is discovered.
Recent discussionsof human and animal aggression have failed to distinguishbetween moralistic and other formsof aggression(e.g., Scott,
1958; Lorenz, 1966; Montague, 1968; Tinber-
ALTRUISMW
49
gen, 1968; Gilula and Daniels, 1969). The
grounds for expecting,on functionalgrounds,
a highlyplastic developmentalsystemaffecting
moralisticaggressionis discussedbelow.
(4) Gratitude,sympathy,and the cost/benefit
ratio of an altruisticact. If the cost/benefit
ratio is an importantparameterin determining
the adaptiveness of reciprocal altruism, then
humansshould be selectedto be sensitiveto the
cost and benefitof an altruisticact, both in
deciding whetherto performone and in deciding whether,or how much, to reciprocate. I
suggestthat the emotion of gratitudehas been
selectedto regulatehuman responseto altruistic
acts and that the emotion is sensitive to the
cost/benefit
ratio of such acts. I suggestfurther
that the emotionof sympathyhas been selected
to motivatealtruisticbehavior as a functionof
the plight of the recipient of such behavior;
crudelyput, the greaterthe potentialbenefitto
the recipient,the greaterthe sympathyand the
more likely the altruistic gesture, even to
strange or disliked individuals. If the recipient's gratitudeis indeed a functionof the cost/
benefitratio, then a sympatheticresponse to
the plight of a disliked individual may result
in considerablereciprocity.
There is good evidence supporting the
psychological importance of the cost/benefit
ratio of altruisticacts. Gouldner (1960) has reviewed the sociologicalliteraturesuggestingthat
the greaterthe need stateof the recipientof an
altruisticact, the greater his tendency to reciprocate; and the scarcerthe resourcesof the
donor of the act, the greater the tendencyof
the recipientto reciprocate. Heider (1958) has
analyzed lay attitudes on altruism and finds
that gratitudeis greatestwhen the altruisticact
does good. Tesser,Gatewood,and Driver (1968)
have shown that American undergraduates
thoughtthey would feel more gratitudewhen
the altruisticact was valuable and cost thebenefactora great deal. Pruitt (1968) has provided
evidence that humans reciprocatemore when
the original act was expensive for the benefactor. He shows that under experimentalconditions more altruismis induced by a gift of
80 per cent of $1.00 than 20 per cent of $4.00.
Aronfreed(1968) has reviewedthe considerable
evidence that sympathymotivatesaltruisticbe-
50
THE QUARTERLY
REVIEW OF BIOLOGY
[VOLUME
46
is based on the fact that many transgressions
performed in private are likely to become
public knowledge. It should often be advantageous to confess sins that are likely to be
discoveredbefore theyactually are, as evidence
of sincerity(see below on detectionof mimics).
(6) Subtle cheating: the evolution of mimics.
Once friendship,moralistic aggression, guilt,
sympathy,and gratitudehave evolved to regulate the altruisticsystem,selection will favor
mimickingthese traitsin order to influencethe
behavior of others to one's own advantage.
Apparent acts of generosityand friendshipmay
induce genuine friendshipand altruismin return. Sham moralisticaggressionwhen no real
cheatinghas occurredmay neverthelessinduce
reparativealtruism. Sham guilt may convincea
wronged friend that one has reformedone's
ways even when the cheating is about to be
resumed. Likewise, selection will favor the
hypocrisyof pretendingone is in dire circumstances in order to induce sympathy-motivated
altruistic behavior. Finally, mimicking sympathy may give the appearance of helping in
order to induce reciprocity,and mimicking
gratitude may mislead an individual into expecting he will be reciprocated. It is worth
emphasizingthat a mimic need not necessarily
Many studieshave supportedthe notion that be conscious of the deception; selection may
whetherintentionalor un- favorfeelinggenuine moralisticaggressioneven
public transgression
intentional,
whetherimmoralor onlysituationally when one has not been wronged if so doing
leads to reparativealtruism(p. 267). leads another to reparativealtruism.
unfortunate,
Instances of the above formsof subtle cheatWallace and Sadalla (1966), for example,
ing are not difficultto find. For typical inshowed experimentallythat individuals who
stances fromthe literatureon hunter-gatherers
brokean expensivemachineweremorelikelyto
see Rasmussen (1931), Balikci (1964), and Lee
volunteerfor a painful experimentthan those
and DeVore (1968). The importanceof these
who did not, but only if theirtransgression
had
formsof cheating can partlybe inferredfrom
been discovered. Investigatorsdisagree on the
the adaptations to detect such cheating disextent to which guilt feelingsare the motivacussed below and from the importance and
tion behind reparative altruism. Epstein and
prevalence of moralistic aggressiononce such
Hornstein (1969) supply some evidence that
cheatingis detected.
guilt is involved, but on the assumption that
(7) Detection of the subtle cheater: trustone feels guilt even when one behaves badly
in private,Wallace and Sadalla's (1966) result worthiness, trust, and suspicion. Selection
contradicts the view that guilt is the only should favor the ability to detect and dismotivatingfactor. That private transgressions criminate against subtle cheaters. Selection
are not as likely as public ones to lead to will clearlyfavordetectingand counteringsham
reparativealtruismis preciselywhat the model moralistic aggression. The argument for the
would predict,and it is possible that the com- others is more complex. Selection may favor
mon psychologicalassumption that one feels distrustingthose who perform altruistic acts
guilt even when one behaves badly in private without the emotional basis of generosityor
havior as a functionof the plight of the individual arousingthe sympathy.
(5) Guilt and reparativealtruism. If an organism has cheated on a reciprocal relationship
and thisfacthas been found out, or has a good
chance of being found out, by the partnerand
if the partnerrespondsby cuttingoffall future
acts of aid, then the cheater will have paid
dearlyforhis misdeed. It will be to thecheater's
advantage to avoid this,and, providingthat the
cheatermakes up forhis misdeed and does not
cheat in the future,it will be to his partner's
benefitto avoid this,since in cuttingofffuture
acts of aid he sacrificesthe benefitsof future
reciprocalhelp. The cheatershould be selected
to make up for his misdeed and to show convincing evidence that he does not plan to
continue his cheating sometimein the future.
In short, he should be selected to make a
reparativegesture. It seems plausible, furthermore, that the emotion of guilt has been selected forin humanspartlyin orderto motivate
the cheater to compensatehis misdeed and to
behave reciprocallyin the future,and thus to
preventthe ruptureof reciprocalrelationships.
Krebs has reviewed the evidence that harming another individual publicly leads to altruisticbehaviorand concludes:
MARCH
1971]
RECIPROCAL
guilt because the altruistictendenciesof such
individuals may be less reliable in the future.
One can imagine, for example, compensating
for a misdeed withoutany emotional basis but
with a calculating,self-serving
motive. Such an
individual should be distrusted because the
calculating spirit that leads this subtle cheater
now to compensate may in the future lead
him to cheat when circumstancesseem more
advantageous(because of unlikelihoodof detection, for example, or because the cheated individual is unlikely to survive). Guilty motivation, in so far as it evidences a more enduring
commitmentto altruism,either because guilt
teaches or because the cheater is unlikely not
to feel the same guilt in the future,seemsmore
reliable. A similar argument can be made
about the trustworthiness
of individuals who
initiate altruistic acts out of a calculating
rather than a generous-hearteddisposition or
who show either false sympathyor false gratitude. Detection on the basis of the underlying
psychological dynamics is only one form of
detection. In many cases, unreliability may
more easily be detected throughexperiencing
the cheater's inconsistentbehavior. And in
some cases,thirdpartyinteractions(as discussed
below) may make an individual's behavior predictable despite underlying cheating motivations.
The anthropologicalliteraturealso abounds
withinstancesof the detectionof subtlecheaters
(see above referencesfor hunter-gatherers).
Although I know of no psychologicalstudies on
the detectionof sham moralisticaggressionand
sham guilt, thereis ample evidence to support
the notion that humans respond to altruistic
acts accordingto theirperceptionof the motives
of the altruist. They tend to respond more
altruisticallywhen they perceive 'the other as
acting "genuinely"altruistic,thatis, voluntarily
dispatchingan altruisticact as an end in itself,
without being directed toward gain (Leeds,
1963; Heider, 1958). Krebs (1970) has reviewed
the literatureon thispoint and notes that help
is more likely to be reciprocated when it is
perceived as voluntary and intentional (e.g.,
Goranson and Berkowitz, 1966; Lerner and
Lichtman, 1968) and when the help is appropriate, that is, when the intentions of the
altruistare not in doubt (e.g., Brehmand Cole,
ALTRUISM
51
1966; Schopler and Thompson, 1968). Krebs
concludes that, "When the legitimacyof apparent altruismis questioned,reciprocityis less
likelyto prevail." Lerner and Lichtman (1968)
have shown experimentallythat those who act
altruisticallyfor ulterior benefit are rated as
unattractiveand are treated selfishly,whereas
those who apparently are genuinely altruistic
are rated as attractiveand are treated altruistically. Berscheid and Walster (1967) have
shown thatchurchwomen tend to make reparationsforharm theyhave committedby choosing
thereparationthatapproximatestheharm (that
is, is neithertoo slightnor too great), presumably to avoid the appearance of inappropriateness.
Rapoport and Dale (1967) have shown that
when two strangers play iterated games of
Prisoner'sDilemma in which the matrix determines profitsfromthe games played there is a
significanttendencyforthe level of cooperation
to drop at the end of the series,reflectingthe
fact that the partnerwill not be able to punish
for "cheating" responses when the series is
over. If a long seriesis brokenup into subseries
with a pause between subseriesfor totalingup
gains and losses,then the tendencyto cheat on
each otherincreasesat the end of each subseries.
These results,as well as some others reported
by Rapoport and Chammah (1965), are suggestive of the instabilitythat exists when two
strangersare consciously tryingto maximize
gain by tradingaltruisticgestures,an instability
thatis presumablyless markedwhen the underlying motivation involves the emotions of
friendship,of likingothers,and of feelingguilt
over harminga friend. Deutsch (1958), for example, has shown that two individuals playing
iterated games of Prisoner's Dilemma will be
more cooperative if a third individual, disliked by both,is present. The perceivedmutual
dislike is presumed to create a bond between
the twoplayers.
It is worthmentioningthat a classicproblem
in social science and philosophy has been
whetherto definealtruismin termsof motives
(e.g., real vs. "calculated" altruism)or in terms
of behavior,regardlessof motive (Krebs, 1970).
This problem reflectsthe fact that, wherever
studied, humans seem to make distinctions
about altruismpartly on the basis of motive,
52
THE QUARTERLY
REVIEW OF BIOLOGY
and this tendency is consistentwith the hypothesis that such discriminationis relevant to
protectingoneself fromcheaters.
(8) Setting up altruisticpartnerships. Selection will favor a mechanism for establishing
reciprocalrelationships.Since humans respond
to acts of altruismwith feelingsof friendship
that lead to reciprocity,one such mechanism
might be the performingof altruistic acts
toward strangers,or even enemies,in order to
induce friendship.In short,do unto othersas
you would have them do unto you.
The mechanismhypothesizedabove leads to
results inconsistentwith the assumption that
humans always act more altruisticallytoward
friendsthan towardothers. Particularlytoward
strangers,humans may initiallyact more altruisticallythan towardfriends.Wright(1942) has
shown, for example, that third grade children
are more likelyto give a more valuable toyto a
strangerthan to a friend. Later, some of these
childrenverballyacknowledgedthat theywere
tryingto make friends.Floyd (1964) has shown
that, after receiving many trinkets from a
friend,humans tend to decrease their giftsin
return,but afterreceivingmanytrinketsfroma
neutral or disliked individual, they tend to
increase their gifts in return. Likewise, after
receiving few trinketsfrom a friend,humans
tend to increase their giftsin return,whereas
receiving few trinketsfrom a neutral or disliked individual resultsin a decrease in giving.
This was interpretedto mean that generous
friendsare taken forgranted(as are stingynonfriends). Generosityfroma non-friendis taken
to be an overtureto friendship,and stinginess
froma friendas evidence of a deterioratingrelationship in need of repair. (Epstein and
Hornstein, 1969, provide new data supporting
this interpretationof Floyd, 1964.)
(9) Multipartyinteractions.In the close-knit
social groups that humans usually live in, selection should favormore complex interactions
than the two-party
interactionsso fardiscussed.
Specifically,selection may favor learning from
the altruisticand cheatingexperiencesof others,
helping others coerce cheaters,formingmultiparty exchange systems,and formulatingrules
for regulated exchanges in such multiparty
systems.
(i) Learning from others. Selection should
[VOLUME
46
favorlearningabout the altruisticand cheating
tendencies of others indirectly,both through
observing interactions of others and, once
linguistic abilities have evolved, by hearing
about such interactionsor hearingcharacterizations of individuals (e.g., "dirty,hypocritical,
dishonest,untrustworthy,
cheatinglouse"). One
important result of this learning is that an
individual may be as concernedabout the attitude of onlookers in an altruisticsituation as
about the attitude of the individual being
dealt with.
(ii) Help in dealing withcheaters. In dealing
with cheaters selection may favor individuals
helping others,kin or non-kin,by direct coercion against the cheateror by everyonerefusing
him reciprocal altruism. One effectof this is
that an individual, throughhis close kin, may
be compensatedfor an altruisticact even after
his death. An individual who dies saving a
friend,for example, may have altruisticacts
performedby the friend to the benefitof his
offspring.Selection will discriminate against
the cheater in this situation, if kin of the
martyr,or others,are willing to punish lack of
reciprocity.
(iii) Generalized altruism. Given learning
from others and multiparty action against
cheaters,selectionmay favora multipartyaltruisticsystemin whichaltruisticacts are dispensed
freelyamong more than two individuals, an
individual being perceived to cheat if in an
altruisticsituation he dispensesless benefitfor
the same cost than would the others,punishment coming not only fromthe other individual in that particular exchange but from the
othersin the system.
(iv) Rules of exchange. Multipartyaltruistic
systemsincrease by several-foldthe cognitive
in detectingimbalancesand deciding
difficulties
whethertheyare due to cheatingor to random
factors. One simplifyingpossibility that language facilitatesis the formulationof rules of
conduct, cheating being detected as infraction
of such a rule. In short,selection may favor
the elaboration of normsof reciprocalconduct.
There is abundant evidence for all of the
above multipartyinteractions(see the above
Thomas (1958),
referenceson hunter-gatherers).
forexample,has shownthatdebtsof reciprocity
do not disappear with the death of the
MARCH
19711
RECIPROCAL
ALTRUISM
53
"creditor"but are extended to his kin. Krebs nism's sense of guilt to be educated, perhaps
has reviewed the psychological literature on partly by kin, so as to permit those formsof
generalized altruism. Several studies (e.g., cheating that local conditions make adaptive
Darlingtonand Macker, 1966) have shown that and to discourage those with more dangerous
humansmay directtheiraltruismto individuals consequences. One would not expect any simother than those who were hurt and may ple systemregulatingthe developmentof altrurespond to an altruisticact that benefitsthem- istic behavior. To be adaptive, altruisticbeselves by acting altruisticallytoward a third havior must be dispensed with regard to many
individual uninvolvedin the initial interaction. characteristicsof the recipient (including his
Berkowitzand Daniels (1964) have shown ex- degree of relationship,emotional makeup, past
perimentally,for example, that help from a behavior, friendships,and kin relations), of
confederate leads the subject to direct more other membersof the group, of the situation
help to a thirdindividual,a highlydependent in whichthe altruisticbehavior takesplace, and
supervisor. Freedman, Wallington, and Bless of many other parameters,and no simple de(1967) have demonstratedthe surprisingresult velopmental system is likely to meet these
that, in two differentexperimentalsituations, requirements.
humans engaged in reparativealtruismonly if
Kohlberg(1963), Bandura and Walters(1963),
it could be directedto someone other than the and Krebs have reviewed the developmental
individualharmed,or to the originalindividual literatureon human altruism.All of themcononly if theydid not expect to meet again. In clude that none of the proposed developmental
a systemof strongmultipartyinteractionsit is theories (all of which rely on simple mechapossible that in some situationsindividuals are nisms) can account for the known diverse deselected to demonstrategeneralized altruistic velopmental data. Whiting and Whiting (in
tendenciesand that their main concern when prep.) have studied altruisticbehavior directed
theyhave harmed anotheris to show that they towardskin by childrenin six different
cultures
are genuinelyaltruistic,which theybest do by and find consistentdifferencesamong the culacting altruisticwithout any apparent ulterior tures that correlate with differencesin childmotive, e.g., in the experiments,by acting rearingand other facetsof the cultures. They
altruistic toward an uninvolved third party. argue that the differences
adapt the childrento
Alternatively,
A. Rapoport (pers. commun.)has differentadult roles available in the cultures.
suggestedthat the reluctance to direct repara- Although the behavior analyzed takes place
tive altruismtowardthe harmedindividualmay between kin and hence Hamilton's model
be due to unwillingness to show thereby a (1964) may apply rather than this model, the
recognitionof the harm done him. The re- Whitings' data provide an instance of the
directionserves to allay guilt feelingswithout adaptive value of developmental plasticityin
triggeringthe greaterreparation that recogni- altruisticbehavior. No careful work has been
done analyzingthe influenceof environmental
tionof theharmmightlead to.
(10) Developmentalplasticity.The conditions factors on the development of altruistic beunder which detection of cheating is possible, havior,but some data exist. Krebs has reviewed
the range of available altruistic trades, the the evidence that altruistictendencies can be
cost/benefitratios of these trades, the relative increased by the effectsof warm, nurturant
stabilityof social groupings,and otherrelevant models, but little is known on how long such
parameters should differfrom one ecological effectsendure. Rosenhan (1967) and Rettig
and social situation to another and should (1956) have shown a correlationbetween altrudifferthroughtime in the same small human ism in parentsand altruismin theircollege-age
population. Under these conditionsone would children, but these studies do not separate
expect selectionto favordevelopmentalplastic- genetic and environmental influences. Class
ity of those traits regulating altruistic and differencesin altruistic behavior (e.g., Berkocheating tendencies and responses to these witz, 1968; Ugurel-Semin,1952; Almond and
tendencies in others. For example, develop- Verba, 1963) may primarily reflect environmental plasticitymay allow the growingorga- mental influences.Finally Lutzker (1960) and
54
THE QUARTERLY
REVIEW OF BIOLOGY
Deutsch (1958) have shownthatone can predict
the degree of altruisticbehavior displayed in
iterated games of Prisoner's Dilemma from
personality typing based on a questionnaire.
are probablypartly
Such personalitydifferences
environmentalin origin.
It is worth emphasizing that some of the
psychologicaltraitsanalyzedabove have applications outside the particularreciprocalaltruistic
systembeing discussed. One may be suspicious,
for example, not only of individuals likely to
cheat on the altruisticsystem,but of any individual likely to harm oneself; one may be suspicious of theknowntendenciestowardadultery
of anothermale or even of these tendenciesin
one's own mate. Likewise, a guilt-motivated
show of reparation may avert the revenge of
someone one has harmed,whetherthat individual was harmed by cheating on the altruistic
systemor in some otherway. And the systemof
reciprocal altruismmay be employed to avert
possible revenge. The Bushmen of the Kalahari, for example, have a saying (Marshall,
1959) to the effectthat,if you wish to sleep with
someone else's wife,you get him to sleep with
yours,then neitherof you goes afterthe other
withpoisoned arrows. Likewise,thereis a large
literatureon the use of reciprocityto cement
friendshipsbetween neighboringgroups, now
engaged in a common enterprise(e.g., Lee and
DeVore, 1968).
The above review of the evidence has only
begun to outline the complexitiesof the human
[VOLUME 46
altruisticsystem.The inherentinstabilityof the
Prisoner'sDilemma, combined with its importance in human evolution,has led to the evolution of a very complex system. For example,
once moralistic aggression has been selected
for to protect against cheating, selection favors sham moralisticaggressionas a new form
of cheating. This should lead to selection for
the abilityto discriminatethe two and to guard
against the latter. The guarding can, in turn,
be used to counter real moralisticaggression:
one can, in effect,impute cheating motives to
another person in order to protectone's own
cheating. And so on. Given the psychological
and cognitive complexity the systemrapidly
acquires, one may wonder to what extent the
importanceof altruismin human evolution set
up a selection pressure for psychologicaland
cognitive powers which partly contributed to
the large increase in hominid brain size during
thePleistocene.
ACKNOWLEDGMENTS
I thankW. H. Drury,E. Mayr,I. Nisbet,E. E.
Williamsand E. 0. Wilson for usefulcomments
on earlierdraftsof thispaper,and I thankespeciallyI. DeVore and W. D. Hamiltonfordetailed
commentand discussion.I thank A. Rapoport
and D. Krebs for access to unpublishedmaterial.
This workwas completedundera NationalScience
Foundationpre-doctoralfellowshipand partially
supported by grant number NIMH 13156 to
I. DeVore.
LIST OF LITERATURE
G. A., and S. VERBA. 1963. The Civic
Culture. Princeton University Press, Princeton,N. J.
J. 1968. Conduct and Conscience.
ARONFREED,
Academic Press,N. Y.
M. 1962. Migration of European
ASHMOLE,
thrushes:a comparativestudy based on ringing
recoveries. Ibis, 104: 522-559.
BAIER, K. 1958. The Moral Point of View. Cornell UniversityPress,Ithaca, N. Y.
A. 1964. Development of basic socioBALIKCI,
economic units in two Eskimo communities.
National Museum of Canada Bulletin No. 202,
Ottawa.
1963. Social
BANDURA, A., and R. H. WALTERS.
Learning and Personality Development. Holt,
Rhinehart and Winston,N. Y.
L. 1968. Responsibility, reciprocity
BERKOWITZ,
ALMOND,
and social distance in help-giving: an experimental investigation of English social class
differences. J. Exp. Soc. Psychol., 4: 66-669.
1964. Affecting
BERKOWITZ, L., and L. DANIELS.
the salience of the social responsibilitynorm:
effectsof past help on the response to dependency relationships. J. Abnorm. Soc. Psychol.,
68: 275-281.
L., and P. FRIEDMAN. 1967. Some
BERKOWITZ,
social class differencesin helping behavior. J.
Personal. Soc. Psychol.,5: 217-225.
1968. Terms, studBERNDT, R., and H. STERNBERG.
ies and experiments on the problems of bird
dispersion. Ibis, 110: 256-269.
1967. When does
BERSCHEID, E., and E. WALSTER.
a harm-doercompensate a victim? J. Personal.
Soc. Psychol.,6: 435-441.
1963. Longevity, palatability and
BLEST, A. D.
MARCH
1971]
RECIPROCAL
natural selection in five species of New World
Saturuiid moth. Nature, 197: 1183-1186.
BREHM,J. W., and A. H. COLE. 1966. Effect of
a favor which reduces freedom. J. Personal.
Soc. Psychol.,3: 420-426.
1965. ExBROWER, J. V. Z., and L. P. BROWER.
perimentalstudiesof mimicry. 8. Am. Natur.,
49: 173-188.
CAMPBELL,
B. 1966. Human Evolution. Aldine,
Chicago.
CROOK, J.H.
1969. The socio-ecologyof primates.
In J.H. Crooke (ed.), Social Behavior in Birds
and Mammals, p. 103-166. Academic Press,
London.
DARLINGTON,
R. B., and C. E. MACKER. 1966. Displacement of guilt-producedaltruisticbehavior.
J. Personal. Soc. Psychol.,4: 442-443.
DARWIN, C. 1871. The Descent of Man and
Selection in Relation to Sex. Random House,
N. Y.
DEUTSCH, M. 1958. Trust and suspicion. J. ConflictResolution, 2: 267-279.
DHONT, A. A., and J. HUBLE'. 1968. Fledging-date
and sex in relation to dispersal in young Great
Tits. Bird Study,15: 127-134.
DRURY,W. H., and W. J. SMITH. 1968. Defense
of feeding areas by adult herring gulls and
intrusionby young. Evolution, 22: 193-201.
EIBL-EIBEsFELDT,
I. 1955. Qber Symbiosen, Parasitismusund andere besondere zwischenartliche
Beziehungen tropischer Meeresfische. Z. f.
7Tierpsychol.,
12: 203-219.
1959. Der Fisch Aspidontus taeniatus als
Nachahmer des Putzers Labroides dimidiatus.
Z. f. Tierpsychol.,16: 19-25.
EPSTEIN, Y. M., and H. A. HORSTEIN. 1969. Penalty
and interpersonalattractionas factorsinfluencing the decision to help another person. J.
Exp. Soc. Psychol.,5: 272-282.
FEDER, H. M. 1966. Cleaning symbioses in the
marine environment. In S. M. Henry (ed.),
Symbiosis,Vol. 1, p. 327-380. Academic Press,
N. Y.
FISHER, R. A. 1958. The Genetical Theory of
Natural Selection. Dover, N. Y.
FLOYD, J. 1964. Effectsof amount of award and
friendshipstatus of the other on the frequency
of sharing in children. Unpublished doctoral
dissertation. University of Minnesota. (Reviewed in Krebs, 1970).
FREEDMAN, J. L., S. A. WALLINGTON, and E. BLESS.
1967. Compliance without pressure: the effect
of guilt. J. Personal. Soc. Psychol., 7: 117-124.
R. W. 1960. Alter versus ego: an
FRIEDRICHS,
exploratory assessment of altruism. Am.
Sociol. Rev., 25: 496-508.
1969. Violence
GILULA, M. F., and D. N. DANIELS.
ALTRUISM
55
and man's struggle to adapt. Science, 164:
395-405.
GORANSON,R., and BERKOWITZ.1966. Reciprocity and responsibilityreactions to prior help.
J. Personal. Soc. Psychol.,3: 227-232.
GOULDNER,
A. 1960. The norm of reciprocity: a
preliminary statement. Am. Sociol. Rev., 47:
73-80.
HALDANE, J. B. S. 1955. Population genetics.
New Biology, 18: 34-51.
HALL, K. R. L., and I. DEVORE. 1965. Baboon
social behavior. In I. DeVore (ed.), Primate
Behavior: Field Studies of Monkeys and Apes,
p. 53-110. Holt, Rhinehart and Winston,N. Y.
HAMILTON, W. D.
1964. The genetical evolution
of social behavior. J. Theoret. Biol., 7: 1-52.
*. 1966. The moulding of senescenceby natural selection. J. Theoret. Biol., 12: 12-45.
-. 1969. Selection of selfish and altruistic behavior in some extreme models. Paper presented at "Man and Beast Symposium" (in
press,SmithsonianInstitution).
1928-1930.
HARTSHORNE, H., and M. A. MAY.
Studies in the Nature of Character. Vol. 1,
Studies in Deceit; Vol. 2, Studies in Self-Control; Vol. 3, Studies in the Organization of
Character. Macmillan, N. Y.
HEDIGER,
H. 1968. Putzer-fische im aquarium.
Natur und Museum, 98:89-96.
HIEIDER, F. 1958. The Psychology of Interpersonal Relations. Wiley,N. Y.
HOBSON, E. S. 1969. Comments on certain recent
generalizationsregarding cleaning symbiosesin
fishes. PacificScience,23: 35-39.
KOHLBERG, L. 1963. Moral development and
identification.In H. W. Stevenson (ed.), Yearbook of the National Society for the Study of
Education. Part 1. Child Psychology,p. 277332. Universityof Chicago Press,Chicago.
KREBS, D. 1970. Altruism-an examination of the
concept and a review of the literature. Psychol. Bull., 73: 258-302.
KUYTON, P. 1962. Verhalten-beobachtungen an
der Raupe des Kaiseratlas. Z. d. Entomol.,
72: 203-207.
LEE, R. 1969. !Kung Bushman violence. Paper
presented at meeting of American Anthropological Association,Nov. 1969.
LEE, R., and I. DEVORE. 1968. Man the Hunter.
Aldine, Chicago.
LEEDS,R. 1963. Altruismand the norm of giving.
Merrill-PalmerQuart., 9: 229-240.
1968. Effects
LERNER,M. J.,and R. R. LICHTMAN.
of perceived norms on attitudes and altruistic
behavior toward a dependent other. J. Personal. Soc. Psychol.,9: 226-232.
THE QUARTERLY
56
REVIEW OF BIOLOGY
1965. Ober die Funktion der relativen Stimmungshierarchie (dasgestellt am
Beispiel der phylogenetischenund ontogenetischen Entwicklungdes Beutefangsvon Raubtieren. Z. f. Tierpsychol.,22: 412-494.
LIMBAUGH, C. 1961. Cleaning symbioses. Scient.
Am., 205: 42-49.
LEYHAUSEN, P.
LIMBAUGH, C., H. PEDERSON, and F. CHASE.
1961.
Shrimps that clean fishes. Bull. Mar. Sci. Gulf
Caribb., 11: 237-257.
LORENZ, K. 1966. On
Aggression. Harcourt,
Brace and World, N. Y.
LUCE, R. D., and H. RAIFFA. 1957. Games and
Decisions. Wiley,N. Y.
LUTZKER,
D. 1960. Internationalismas a predictor
of cooperative game behavior. J. Conflict
Resolution, 4: 426-435.
MARLER,P. 1955. The characteristicsof certain
animal calls. Nature, 176: 6-7.
. 1957. Specific distinctiveness in the communication signals of birds. Behavior, 11:
13-39.
MARSHALL, L.
K.
1959.
Marriage
among
!Kung
Bushmen. Africa,29: 335-365.
. 1961. Sharing, talking and giving: relief of
social tension among !Kung Bushmen. Africa,
31: 231-249.
E. C. L. 1968. Cleaning symbiosisand
MAYNARD,
oral grooming on the coral reef. In P. Person
(ed.), Biology of the Month, p. 79-88. Philip
Person, American Association for the Advancement of Science,Wash., D. C.
MAYNARDSMITH, J.
1964.
Kin selection and group
selection. Nature, 201: 1145-1147.
. 1965. The evolution of alarm calls. Amer.
Natur., 99: 59-63.
MAYR, E. 1963. Animial Species and Evolution.
Belknap Press,Cambridge.
MONTAGU, F. M. A. 1968. Man and Aggression.
Oxford UniversityPress,N. Y.
MURIE, A. 1944. The Wolves of Mount McKinley.
Fauna of National Parks, Faunal Series # 5;
Wash., D.C.
OWEN,D. F. 1963. Similar polymorphismasin an
insect and a land snail. Nature, 198: 201-203.
PERDECK,A. 1958. Two types of orientation in
migrating starlings, Starnus vulgaris L., and
chaffishes,Fringilla coeleus L., as revealed by
displacement experiments. Ardea, 46: 1-35.
PRUITT,D. G. 1968. Reciprocityand credit building in a laboratory dyad. J. Personal. Soc.
Psychol., 8: 143-147.
RANDALL,
J. E. 1958. A review of the Labrid fish
genus Labroides with descriptionsof two new
species and notes on ecology. Pacific Science,
12: 327-347.
[VOLUME
46
. 1962. Fish service stations. Sea Frontiers,
8: 40-47.
RAPOPORT,
A., and A. CHAMMAH. 1965. Prisoner's
Dilemma. Universityof Michigan Press, Ann
Arbor.
RAPOPORT,
A., and P. DALE. 1967. The "end" and
"start" effectsin iterated Prisoner's Dilemma.
J.ConflictResolution, 10: 363-366.
RASMUSSEN,K. 1931. The Netsilik Eskimos: social
life and spiritual culture. Report of the Fifth
Thule Expedition 1921-1924, Vol. 8(1,2).
GyldendalskeBoghandel, Copenhagen.
RETTIG, S. 1956. An exploratorystudyof altruism.
Dissert.Abstr.,16: 2220-2230.
REYNOLDS, L. 1966. Open groups in hominid
evolution. Man, 1: 441-452.
ROSENHAN, D. 1967. The Origins of Altruistic
Social Autonomy. Educational Testing Service, Princeton,N.J.
RoussEAu,J. J. 1954. The Social Contract. Henry
RegneryCo., Chicago.
RUDEBECK, G. 1950. The choice of prey and
modes of hunting of predatorybirds with special reference to their selective effort. Oikos,
2: 65-88.
* 1951. The choice of prey and modes of
hunting of predatorybirds, with special reference to their selective effort(cont.) Oikos, 3:
200-231.
SAWYER, J. 1966. The altruism scale: a measure
of cooperative, individualistic,and competitive
interpersonal orientation. Am. J. Social. 71:
407-416.
SCHOPLER,J.,and V. T. THOMPSON. 1968. The role
of attributionprocess in mediating amount of
reciprocityfor a favor. J. Personal. Soc. Psychol., 10: 243-250.
SCOTT, J. P. 1958. Aggression.Universityof Chicago Press,Chicago.
SOUTHERN, H. N. 1954. Tawny owls and their
prey. Ibis, 96: 384-410.
STRUHSAKER, T. 1967. Social structure among
vervet monkeys (Cercopithecus aethiops). Behavior,29: 83-121.
TESSER, A., R. GATEWOOD, and M. DRIVER. 1968.
Some determinants of gratitude. J. Personal.
Soc. Psychol.,9:\232-236.
THOMAS, E. M. 1958. The Harmless People.
Random House, N. Y.
TINBERGEN, L. 1960. The dynamics of insect and
bird populations in pine woods. 1. Factors
influencingthe intensityof predation by song
birds,Arch. Neerl. de Zool., 13: 265-343.
TINBERGEN, N. 1968. On war and peace in animals and man. Science, 160: 1411-1418.
UGUREL-SEMIN, R. 1952. Moral behavior and
MARCH
1971]
RECIPROCAL
ALTRUISM
57
moral judgment of children. J. Abnorm. Soc. WICKLER,W. 1968. Mimicry in Plants and AniPsychol.,47: 463-474.
mals. McGraw-Hill,N. Y.
J. 1968. A
preliminary WILIIAMS, G. C. 1966. Adaptation and Natural
VANLAWICK-GoODALL,
report on expressivemovementsand communiSelection. Princeton UniversityPress, Princecation in the Gombe Stream chimpanzees. In
ton,N. J.
P. Jay (ed.), Primates, p. 313-374 Holt, RhineWRIGHT,B. 1942. Altruism in children and the
hart and Winston, N.Y.
perceived conduct of others. J. Abnorm. Soc.
1966. Behavioral
WALLACE, J., and E. SADALLA.
Psychol.,37: 218-233.
consequences of transgression: the effectsof
WYNNE-EDWARDS,
V. C. 1962. Animal Dispersion
1:
Personal.,
Exp.
Res.
social recognition. J.
in Relation to Social Behavior. Hafner, N. Y.
187-194.